AMD is not Intel

AMD (AMD) is now valued at nearly 1x Intel’s (INTC) value. Here, I will put this valuation to the test. Specifically, I will take a long-term stance to determine whether they can either expand their valuation (in Intel’s case), or grow into it (in AMD’s case).

As the title already revealed, I do not believe that AMD could ever really become the size of Intel. Although on the surface, Intel and AMD may look like similar companies with similar products, ultimately Intel’s scope is just so much wider, covering 5G, IoT, the foundry market and autonomous driving. All of those are important segments which AMD has no investments in.

As such, the current relative valuation of both companies seems untenable in the long term. Although there is a likelihood that AMD could grow into its valuation, for the relative valuation to become more reflective of the underlying reality, I mainly see Intel growing much larger. This is because its valuation has become so compressed given the failures of prior management.

Background
Over a year ago, when AMD reached the $100B mark, I claimed that AMD was not half an Intel as the stock had quickly surged from the $50s to $80s (for a large part due to Intel’s 7nm delay). This article will update that evaluation.

Nevertheless, earlier this year, after AMD traded sideways for nearly a year, I called AMD undervalued. The stock has rallied 50% since that call.

AMD thesis: Flawed and limited
Let’s start with the AMD thesis. This is actually the easiest part (compared to the financials/valuation). Clearly, the main thesis for AMD remains to take market share from Nvidia (NVDA) in graphics and Intel in CPUs. As I wrote in May:

However, what most will really care about remains unknown. How much Ryzen did AMD sell? How much Radeon? What is the split between console and Epyc sales? AMD doesn’t want investors to know.

CPU: $30B revenue target
On the CPU side, my outlook is neutral to slightly bearish. As I have discussed previously, Intel’s upcoming portfolio in 2022 will be such a major upgrade (Intel calls it its biggest update in a decade) that I previously argued that the AMD threat for Intel would be finished going forward. Additionally, as Intel said on the recent Q3 call, in these times of shortages, capacity is destiny, and I regard Intel as better-positioned to provide supply than AMD, which is reliant on TSMC (TSM).

However, a few details prevent me from becoming bearish just yet. Although Intel’s PC roadmap is rock solid, I remain puzzled by Intel’s execution and progress in the data center. (For example, Meteor Lake was delayed by about 1-2 quarters compared to Granite Rapids’ delay of well over a year.) Additionally, although it has taken several years, by now, one has to recognize that AMD has obtained critical momentum and is seen as a legitimate alternative to Intel. Since even now the majority of Intel’s data center revenue still comes from 14nm parts, this momentum may continue for a while.

Hence, I see no reason to doubt Pat Gelsinger’s statement that he is targeting 2024-2025 before the results of the new investments will become visible. In that longer-term timeline, the current shortages will also be relegated to just a temporary note in history.

In the neutral case, I would expect AMD’s market share to steadily increase to ~30%. If we assume the PC + data center TAM grows to $100B, then AMD could reach $30B in revenue over time. In the worst-case, Intel regaining leadership could mean that AMD may again become the value brand.

GPU: $10B revenue target
On the GPU side, I am slightly more bullish since I see Nvidia’s market share position as untenable. AMD’s latest accelerator seems about as powerful as Intel’s upcoming 100 billion transistor 5nm Ponte Vecchio behemoth (which is over 2x Nvidia’s A100 in theoretical performance).

However, one more invisible issue in the data center may be software (optimizations). Nevertheless, I do not quite subscribe to the view that CUDA is the holy grail, since, for example, AI is being done with vendor-agnostic frameworks such as TensorFlow. Still, taking market share from Nvidia may prove to be difficult after all.

Conclusion
Overall, then, most would assume that AMD’s growth markets such as gaming and data center will continue to grow in the long term, and some would argue AMD could grow faster than the overall market as well. Combining the above CPU estimate with consoles and graphics, then AMD may perhaps a bit more than double its revenue to $40B over time. If we assume AMD reaches this level in 2030, then it would be about two to three decades behind Intel.

Intel thesis: Undervalued and underestimated
In the last several years, although its financials have been decent, according to general consensus, Intel has fallen behind in technology and now trades at a measly single-digit P/E. However, as I see it, Intel is going nowhere. And this is especially true now that Intel finally has a capable CEO that has gotten the greenlight from the board to throw all its funding at innovation (rather than stock buybacks). And this is exactly what Intel will be doing. As I recently argued, the stock market generally does not value a company based on its roadmap, which means investors have ample time to get on board.

CPU
Indeed, these roadmaps imply Intel could legitimately turn around in the next three to four years. As argued in previous section, Intel seems pretty safe on the PC side. In the data center, Intel has hinted that Granite Rapids (scheduled for 2023 after the aforementioned delay) will contain 2x as many cores as Sapphire Rapids, which means that Intel could leapfrog AMD (which will have a 96-core CPU then) with a 112-core CPU.

In summary, Intel will get its roadmap back on track over the next few years, and will hence maintain its market share going forward, well before AMD could get anywhere near the 40% or more market share that AMD really needs to justify its valuation.

Other
However, the main reason Intel could deserve a premium valuation is because it is targeting a market that is simply so much larger than AMD’s. In the article where I claimed AMD is not half an Intel, I described Intel as a semiconductor conglomerate. In the past, I have discussed 5G, robotaxis and the foundry business. Intel also has some smaller growth businesses with Optane memory and IoT, which grew over 50% last quarter to a new record.

I have previously estimated Mobileye could become as large as the data center or PC, for example. Pat Gelsinger also added an additional wildcard with the Intel Foundry Services business. I have been closely monitoring progress in Intel and TSMC’s roadmaps, and in the most bullish case, Intel could start to take significant market share if it really achieves its target to overtake TSMC by 2025. This would be similar to how AMD has managed to substantially increase its revenue in the last few years by taking market share from Intel; in this case, Intel might start to take significant share (and hence revenue) from TSMC.

As such, Intel has exposure to all major growth areas in semiconductors: cloud, AI, graphics, 5G, IoT, foundry and autonomous driving. This means Intel competes in about 2x as many markets as AMD. Hence, to approximation, AMD cannot be worth even half as much as Intel (without having >50% market share).

Conclusion
For some more concrete numbers, Intel discussed its long-term growth target at the recent Q3 call (10-12%). Intel was a bit ambiguous about when it would start to grow at that rate, so for simplicity, I will assume that Intel will be $80B in 2025 and then grow at a 5-10% CAGR through 2030. In that scenario, Intel will generate $100-130B by 2030. This is 2.5x to 3.3x more than my estimate for AMD.

Case study: Intel first to 3nm
As I have discussed in previous articles, Intel has been rumored since 2020 to be one of the first to transition to TSMC (TSM) 3nm. The latest rumor has provided further details about this: Intel would produce the GPU tile in its Meteor Lake CPU at TSMC.

Valuation
For AMD, the best-case seems about priced in, trading at a ~50x P/E, despite that going forward AMD will not hold any significant/differentiated technology proposition anymore (as was the case when Intel was stuck on 14nm).

Meanwhile, Intel used its recent Q3 to provide investors with the financial reality of Pat Gelsinger’s comeback plan, which will see gross margin dip to 51-53%. This has led the stock price to sink to single-digit P/E. As such, one could claim that Intel is currently being priced for the worst-case scenario: Intel will be spending a ton (over $40B annually in R&D and capex), and the market does not believe Intel will see any returns from these investments.

Nevertheless, in the best-case, Intel will continue to steadily increase revenue from its current businesses (as it has done from 2015-2020) while building significant new businesses such as Xe GPUs, MoovitAV (robotaxis), Mobileye Drive (self-driving consumer vehicles) and Intel Foundry Services. Additionally, as Intel executes to regain industry technology leadership, then it will again be able to get back to industry-leading ASPs, resulting in both revenue and gross margin upside.

Of course, the proof would ultimately be in the pudding: Intel would have to show the market (1) that it can again execute reliably on its leadership process roadmap and hence get its gross margin back up, and (2) that it can grow its revenue at the expense of its competitors (AMD, Nvidia, TSMC). To that end, the Qualcomm foundry win for Intel’s 20A node may be seen as a small first step. Nevertheless, for now, the most promising sign may be the amount of veterans that have joined or rejoined Intel in the wake of Pat Gelsinger CEO appointment.

To put this in some more numbers, in previous sections, I estimated AMD could reach $40B by 2030 compared to Intel’s ~$120B. Under the assumption that, a decade from now, nothing will be visible anymore from Intel’s current issues, one could argue that Intel, by virtue of being an IDM, most likely will have improved its margins back to ~60% (higher than AMD).

As such, one could argue that Intel should actually be valued higher than AMD. In that case, the relative valuation between AMD and Intel might expand (in favor of Intel) from the current 1x to (at least) 3x (based on 3x the revenue). This would result in a market cap for Intel of up to $1 trillion. Of course, this is under the assumption that there would be no further missteps in the business that would continue to deflate Intel’s valuation like it currently is.

In the table below, the AMD valuation is based on either AMD growing into its current valuation, or further expanding over time in line with revenue.

Lastly, for comparison, Tesla (TSLA) has now grown to a record $1 trillion valuation, despite FSD still being merely L2 and in beta while Intel is starting its L4 robotaxi business next year. Tesla also has lower revenue and 20 points lower gross margin. As discussed previously, Intel seems better positioned for autonomous driving than Tesla: Uber, Intel Will Dominate Autonomous Driving. Legions of investors have ascribed enormous valuations to Tesla’s non-existing robotaxi business, so from that view, a $1 trillion Intel wouldn’t be all that far-fetched.

The tide is turning
The tide seems to be turning among a wider range of investors and analysts, as discussed for example on Seeking Alpha:

Alder Lake
I had discussed Alder Lake in many articles before its launch, and now the reviews are available. Most investors doubted that Alder Lake would be game-changing for Intel: Intel Stock: Alder Lake Looks Like a Game-Changer (NASDAQ:INTC). However, the reviews are completely unambiguous: Intel is back. Especially, the Core i5 has received universal praise (vs. Ryzen 5).

When averaging out dozens of benchmarks, it can be seen that each Alder Lake CPU is faster than the one it competes against (note, for example, the Core i5 being nearly on-par with Ryzen 7):

Alder Lake’s strongest performance:

Source: Tweakers

Investor takeaway
In my recent discussion of Intel’s Q3 results, I compared Intel a few times against AMD, as I see Intel capable of a similar comeback (even on a bigger scale), and I elaborated on that topic here. To be specific, given Intel’s issues in the recent past, the company has become the underdog, similar to AMD just half a decade ago.

Nevertheless, by revenue and net income, Intel is still by far the world’s biggest logic semiconductor company. This means Intel has ample resources to get itself back together, and it currently also has a CEO that is doing exactly that. If a near-bankrupt AMD could become what it is today (nearly Intel’s valuation), then investors might perhaps give a second thought about what Intel could become if it finally starts to turn that combined $40B annual R&D and capex budget (more than double AMD’s annual revenue) into results.

Even when using conservative growth estimates for Intel and generous ones for AMD, in this article, I arrived at the conclusion that Intel in 2030 should still be about 3x larger than AMD. Hence, as the title stated, AMD is not three-quarters Intel. If one would then translate that difference into market cap, then Intel could become well over $1 trillion in market cap. Seeking Alpha